Profile and Characterization of the Chlorogenic Acids in Green Robusta Coffee Beans by LCMS n :...

pubs.acs.org/JAFC Published on Web 07/20/2010 © 2010 American Chemical Society

8722 J. Agric. Food Chem. 2010, 58, 8722–8737

DOI:10.1021/jf1014457

Profile and Characterization of the Chlorogenic Acids in GreenRobusta Coffee Beans by LC-MSn: Identification of Seven

New Classes of Compounds

RAKESH JAISWAL, MARIA ALEXANDRA PATRAS, PINKIE JACOB ERAVUCHIRA, AND

NIKOLAI KUHNERT*

School of Engineering and Science, Chemistry, Jacobs University Bremen, 28759 Bremen, Germany

LC-MSn (n = 2-4) has been used to detect and characterize in green Robusta coffee beans 15

quantitatively minor sinapic acid and trimethoxycinnamoylquinic acid-containing chlorogenic acids,

all reported for the first time from this source, with 13 of them not previously reported in nature. These

comprise 3-sinapoylquinic acid, 4-sinapoylquinic acid, and 5-sinapoylquinic acid (Mr 398); 3-sinapoyl-5-

caffeoylquinic acid, 3-sinapoyl-4-caffeoylquinic acid, and 4-sinapoyl-3-caffeoylquinic acid (Mr 560);

3-(3,5-dihydroxy-4-methoxy)cinnamoyl-4-feruloylquinic acid (Mr 560); 3-sinapoyl-5-feruloylquinic acid,

3-feruloyl-4-sinapoylquinic acid, and 4-sinapoyl-5-feruloylquinic acid (Mr 574); 4-trimethoxycinnamoyl-5-

caffeoylquinic acid, 3-trimethoxycinnamoyl-5-caffeoylquinic acid (Mr 574); and 5-feruloyl-3-trimethoxy-

cinnamoylquinic acid, 3-trimethoxycinnamoyl-4-feruloylquinic acid, and 4-trimethoxycinnamoyl-5-feruloyl-

quinic acid (Mr 588). Furthermore, a series of structures including nine new triacyl quinic acids have

been assigned on the basis of LC-MSn patterns of fragmentation, relative hydrophobicity, and analogy

of fragmentation patterns if compared to feruloyl, caffeoyl, and dimethoxycinnamoyl quinic acids. Sixty-

nine chlorogenic acids have now been characterized in green Robusta coffee beans.

KEYWORDS: Chlorogenic acids; coffee; sinapoylquinic acids; sinapoyl-caffeoylquinic acids; sinapoyl-feruloylquinic acids; trimethoxycinnamoylquinic acids; caffeoyl-trimethoxycinnamoylquinic acids; feruloyl-trimethoxycinnamoylquinic acids; dimethoxy-feruloyl-caffeoylquinic acid; triacyl quinic acids; LC-MSn

INTRODUCTION

Classically, chlorogenic acids are a family of esters formedbetween quinic acid and certain trans-cinnamic acids, mostcommonly caffeic, p-coumaric, and ferulic acid (1-3). Represen-tative structures are shown in Figure 1. In the IUPAC system(-)-quinic acid is defined as 1 L-1(OH),3,4/5-tetrahydroxycyclo-hexane carboxylic acid, but Eliel and Ramirez (4) recommend1R,3R,4R,5R-tetrahydroxycyclohexane carboxylic acid. Chloro-genic acids are widely distributed in plants (2, 3), but the coffeebean is remarkably rich, containing at least 45 chlorogenic acidsthat are not acylated at C1 of the quinic acid moiety. In previouswork we have profiled the chlorogenic acids in green Arabicacoffee beans and identified a total of 45 different chlorogenicacids (5-8).

The aim of this studywas to profile greenRobusta coffee beanswith respect to their chlorogenic acid profile. Coffee trees areusually divided botanically into two categories, Coffea arabica(Arabica coffee) and Coffea canephora (usually referred to asRobusta coffee). Supposedly high-quality coffee blends consisttypically of 100% Arabica coffee beans. Lower quality, cheaperblendsmay have some proportion of Robusta beans, or theymayconsist entirely of Robusta. Arabica beans produce allegedly a

superior taste in the cup, being more flavorful and complex thantheir Robusta counterparts. Robusta beans in contrast tend toproduce amore bitter brew, with amusty flavor and less body (9).Obviously this difference in sensory properties could be related tothe individual phytochemical profile of the two coffee varieties.Because chlorogenic acids are themost common secondarymeta-bolites found in the green coffee bean and are therefore precursorsfor the formation of further compounds with sensory propertiesin the coffee roasting process, a closer inspection of their chloro-genic acid profile deserves attention to allow a direct comparisonof compounds present in either coffee variety. Hofmann and co-workers (9), for example, have shown that chlorogenic acids andtheir derivatives contribute considerably to the sensory propertiesof the coffee brew.

The cultivation of Robusta coffee if compared to Arabica hasseveral advantages because their trees produce their first cropswithin about 2-3 years of being planted. Arabica trees, in com-parison, require about 4-5 years to yield fruit. Farmers havehence a good motive to grow the faster growing variety and takeadvantage of upswings in the price of coffee. Additionally, theRobusta coffee tree can grow under a larger variety of climaticconditions than can the Arabica because it is in general moretolerant to cold climate and tolerates a wider range of altitudes.

Recently, LC-MSn has been used to characterize cinnamoyl-amino acid conjugates (10) and to discrimate between indivi-dual isomers of monoacyl and diacyl chlorogenic acids (5-8).

*Author to whom correspondence should be addressed (phone49 421 200 3120; fax 49 421 200 3229; e-mail [email protected]).

Article J. Agric. Food Chem., Vol. 58, No. 15, 2010 8723

Figure 1. Continued

8724 J. Agric. Food Chem., Vol. 58, No. 15, 2010 Jaiswal et al.

Figure 1. Continued


Figure 1. Continued


The MS fragmentation patterns in tandem MS spectra, UVspectrum, retention time, and relative hydrophobicity have beenutilized to develop structure-diagnostic hierarchical keys for theidentification of chlorogenic acids. In this study, we applied thesemethods to the qualitative profiling of chlorogenic acids in greenRobusta coffee bean extract.

MATERIALS AND METHODS

All of the chemicals (analytical grade) were purchased from Sigma-Aldrich (Bremen, Germany). Green Robusta coffee beans (Cameroon,India, Indonesia, Tanzania, Togo, and Uganda Robusta) were purchasedfrom supermarkets in Bremen, Germany.

Methanolic Extract of Coffee Beans. Six different samples of greenRobusta coffee (Cameroon, India, Indonesia, Tanzania, Togo, andUganda Robusta) beans (5 g of each) were freeze-dried overnight at-20 �C and ground to fine powder; methanolic extracts were prepared bySoxhlet extraction using aqueous methanol (70%) for 5 h. The extract wastreated with Carrez reagents (1 mL of reagent A plus 1 mL of reagent B)(11) to precipitate colloidal material and filtered through aWhatman no. 1filter paper. The methanol and water were removed in vacuo, and theresidue was stored at-20 �C until required, thawed at room temperature,dissolved in methanol (60 mg/10 mL), filtered through a membrane filter,and used for LC-MS.

LC-MSn. The LC equipment (Agilent 1100 series, Karlsruhe,

Germany) comprised a binary pump, an autosampler with a 100 μL loop,and a DAD detector with a light-pipe flow cell (recording at 320 and254 nmand scanning from200 to 600 nm). Thiswas interfacedwith an ion-trap mass spectrometer fitted with an ESI source (Bruker Daltonics HCTUltra, Bremen,Germany) operating in full scan, autoMSnmode to obtainfragment ion m/z. As necessary, MS2, MS3, and MS4 fragment-targetedexperiments were performed to focus only on compounds producing aparent ion at m/z 397, 559, 573, 587, 677, 691, 705, or 719. Tandem massspectra were acquired in auto-MSn mode (smart fragmentation) using aramping of the collision energy. Maximum fragmentation amplitude wasset to 1 V, starting at 30% and ending at 200%. MS operating conditions(negative mode) had been optimized using 5-caffeoylquinic acid with acapillary temperature of 365 �C, a dry gas flow rate of 10 L/min, and anebulizer pressure of 10 psi. High-resolution LC-MSwas carried out usingthe same HPLC equipped with a MicrOTOF Focus mass spectrometer(Bruker Daltonics, Bremen, Germany) fitted with an ESI source, andinternal calibration was achieved with 10 mL of 0.1 M sodium formatesolution injected through a six-port valve prior to each chromatographicrun. Calibration was carried out using the enhanced quadratic mode.

HPLC. Separation was achieved on a 150 � 3 mm i.d. column con-taining diphenyl 5 μm, with a 5 mm � 3 mm i.d. guard column (Varian,Darmstadt, Germany). Solvent A was water/formic acid (1000:0.005 v/v)and solvent B was methanol. Solvents were delivered at a total flow rate of500 μL/min. The gradient profile was from 10 to 70%B linearly in 60 minfollowed by 10 min isocratic and a return to 10% B at 90 and 10 minisocratic to re-equilibrate.

Synthesis of Mixture of Regioisomers of Sinapoylquinic Acid

(13-15 and 70). To a solution of quinic acid (100 mg, 0.52 mmol) and

DMAP (8mg, 0.06 mmol) in CH2Cl2 (10mL) were added pyridine (4 mL)and 4-acetylsinapic acid chloride (147 mg, 0.52 mmol) at room tempera-ture. The reaction mixture was stirred for 12 h and acidified with 1MHCl(pH ≈3). The layers were separated, and the aqueous phase was re-extracted with CH2Cl2 (3�20 mL). The combined organic layers weredried over Na2SO4 and filtered, and the solvents were removed in vacuo.The resulting crude esters were dissolved in a mixture of 20 mL oftrifluoroacetic acid and water (7:3) at room temperature and stirred for30 min. The solvents were removed in vacuo, and the resulting reddishproduct was analyzed by HPLC-MS.

Synthesis of Mixture of Regioisomers of (3,4,5-Trimethoxycin-

namoyl)quinic Acid (71-74). 3,4,5-Trimethoxycinnamic acid (500 mg,2.1 mmol) was suspended in toluene (15 mL) containing 3 drops of DMF,and oxalyl chloride (480 mg, 3.78 mmol) was added at 0 �C. The sus-pension was stirred for 10 h at room temperature, and a clear yellowishsolutionwas formed. Toluene and unreacted oxalyl chloridewere removedin vacuo. The yellow residuewaswashedwith petroleum ether and dried invacuo to give 3,4,5-trimethoxycinnamic acid chloride (511mg, 1.99mmol,95%). To a solution of quinic acid (100mg, 0.52mmol) andDMAP (8mg,0.06 mmol) in CH2Cl2 (10 mL) were added pyridine (4 mL) and 3,4,5-trimethoxycinnamic acid chloride (134 mg, 0.52 mmol) at room tempera-ture. The reaction mixture was stirred for 12 h at room temperature andacidified with 1MHCl (pH≈3). The layers were separated, and the aque-ous phase was re-extracted with CH2Cl2 (3� 20 mL). The combinedorganic layers were dried over Na2SO4 and filtered, and the solvents wereremoved in vacuo. The resulting esters were analyzed by HPLC.

RESULTS AND DISCUSSION

Preliminary Assessment of Data. All data for chlorogenic acidspresented in this paper use the recommended IUPAC numberingsystem (1), and structures are presented in Figure 1. When neces-sary, previously published data have been amended to ensureconsistency and avoid ambiguity. In this study a selection of sixdifferent samples of green Robusta coffee beans were analyzedcovering a selection of growing areas and countries of origin.Chlorogenic acids were isolated as described previously (8) usinga Soxhlet extraction followed by protein precipitation andremoval.

In a first set of experiments the Robusta coffee extracts wereanalyzed using a slightly modified HPLC method from the onepreviously reported, in which the stationary phase phenylhexylpacking was exchanged for a diphenyl packing and in the mobilephase acetonitrile was replacedwithmethanol. Using these modi-fied conditions chromatograms were produced using a high-resolution TOF mass detector. The Robusta coffee extract gavea typical chromatogram, in which the 37 chlorogenic acids pre-viously reported in green Arabica beans were readily locatedaccording to theirm/z ratio (5). All 37 chlorogenic acids shown inTable 1 could be identified according to their high-resolutionmass data in the negative ion mode and using tandem LC-MS.

Figure 1. Structures of green coffee bean chlorogenic acids (IUPAC numbering) (1).


The mass error was typically under 5 ppm, confirming themolecular formulas of the chlorogenic acids fromRobusta coffee

previously assigned (5-8,12). All 37 previously reported chloro-genic acids could be readily identified in all five samples of green

Table 1. High-Resolution Mass (MS-TOF) Data of Chlorogenic Acids and Their Parent Ions (M - H)

no. name abbrev mol formula theor m/z (M - H) exptl m/z (M - H) error (ppm)

1 3-O-caffeoylquinic acid 3-CQA C16H18O9 353.0878 353.0881 -0.7



4 3-O-feruloylquinic acid 3-FQA C17H20O9 367.0929 367.1047 -3.4

5 4-O-feruloylquinic acid 4-FQA C17H20O9 367.0929 367.1038 -0.8

6 5-O-feruloylquinic acid 5-FQA) C17H20O9 367.0929 367.1045 -2.9

7 3-O-p-coumaroylquinic acid 3-pCoQA C16H18O8 337.0929 337.0931 -0.5

8 4-O-p-coumaroylquinic acid 4-pCoQA C16H18O8 337.0929 337.0921 2.4

9 5-O-p-coumaroylquinic acid 5-pCoQA C16H18O8 337.0929 337.0921 2.4

10 3-O-dimethoxycinnamoylquinic acid 3-DQA C18H22O9 381.1191 381.1202 -2.8



13 3-O-sinapoylquinic acid 3-SiQA C18H22O10 397.1140 397.1125 3.8

14 4-O-sinapoylquinic acid 4-SiQA C18H22O10 397.1140 397.1150 -2.5

15 5-O-sinapoylquinic acid 5-SiQA C18H22O10 397.1140 397.1140 -4.9

16 3,4-di-O-caffeoylquinic acid 3,4-diCQA C25H24O12 515.1195 515.1190 1.0



19 3,4-di-O-feruloylquinic acid 3,4-diFQA C27H28O12 543.1508 543.1512 -0.8



25 3-O-feruloyl-4-O-caffeoylquinic acid 3F-4CQA C26H26O12 529.1351 529.1343 1.7

26 3-O-caffeoyl-4-O-feruloylquinic acid 3C-4FQA C26H26O12 529.1351 529.1351 -0.1

27 3-O-feruloyl-5-O-caffeoylquinic acid 3F-5CQA C26H26O12 529.1351 529.1373 -4.0

28 3-O-caffeoyl-5-O-feruloylquinic acid 3C-5FQA C26H26O12 529.1351 529.1367 -3.0

29 4-O-feruloyl-5-O-caffeoylquinic acid 4F-5CQA C26H26O12 529.1351 529.1351 0.1

30 4-O-caffeoyl-5-O-feruloylquinic acid 4C-5FQA C26H26O12 529.1351 529.1349 0.5

31 3-O-dimethoxycinnamoyl-4-O-caffeoylquinic acid 3D-4CQA C27H28O12 543.1508 543.1488 3.6

32 3-O-dimethoxycinnamoyl-5-O-caffeoylquinic acid 3D-5CQA C27H28O12 543.1508 543.1491 3.1

33 4-O-dimethoxycinnamoyl-5-O-caffeoylquinic acid 4D-5CQA C27H28O12 543.1508 543.1526 -3.4

34 3-O-dimethoxycinnamoyl-4-O-feruloylquinic acid 3D-4FQA C27H28O12 543.1508 543.1508 -4.1



37 3-O-p-coumaroyl-4-O-caffeoylquinic acid 3pCo-4CQA C25H24O11 499.1246 499.1227 3.7

38 3-O-caffeoyl-4-O-p-coumaroylquinic acid 3C-4pCoQA C25H24O11 499.1246 499.1247 -0.2

39 3-O-p-coumaroyl-5-O-caffeoylquinic acid 3pCo-5CQA C25H24O11 499.1246 499.1248 -0.5



42 4-O-p-coumaroyl-5-O-caffeoylquinic acid 4pCo-5CQA C25H24O11 499.1246 499.1249 -0.6

43 3-O-p-coumaroyl-4-O-feruloylquinic acid 3pCo-4FQA C26H26O11 513.1402 513.1389 2.6

44 3-O-p-coumaroyl-5-O-feruloylquinic acid 3pCo-5FQA C26H26O11 513.1402 513.1141 -2.9

45 4-O-p-coumaroyl-5-O-feruloylquinic acid 4pCo-5FQA C26H26O11 513.1402 513.1406 -0.7

49 3-O-sinapoyl-5-O-caffeoylquinic acid 3Si-5CQA C27H28O13 559.1457 559.1481 -4.2

50 3-O-sinapoyl-4-O-caffeoylquinic acid 3Si-4CQA C27H28O13 559.1457 559.1472 -2.6

51 3-O-(3,5-dihydroxy-4-methoxy)cinnamoyl-4-O-feruloylquinic acid 3DM-4FQA C27H28O13 559.1457 559.1458 -0.2

52 4-O-sinapoyl-3-O-caffeoylquinic acid 4Si-3CQA C27H28O13 559.1457 559.1457 0.9

53 3-O-sinapoyl-5-O-feruloylquinic acid 3Si-5FQA C28H30O13 573.1614 573.1641 -4.7



56 4-O-trimethoxycinnamoyl-5-O-caffeoylquinic acid 4T-5CQA C28H30O13 573.1614 573.1611 0.4

57 3-O-trimethoxycinnamoyl-5-O-caffeoylquinic acid 3T-5CQA C28H30O13 573.1614 573.1623 -1.7

58 3-O-trimethoxycinnamoyl-5-O-feruloylquinic acid 3T-5FQA C29H32O13 587.1770 587.1748 3.8



61 3-O-dimethoxycinnamoyl-4-O-feruloyl-5-O-caffeoylquinic acid 3D-4F-5CQA C37H36O15 719.1981 719.2001 -2.7

62 3,4,5-tri-O-caffeoylquinic acid 3,4,5-triCQA C34H29O15 677.1512 677.1522 -3.5

63 3,5-di-O-caffeoyl-4-O-feruloylquinic acid 3,5-diC-4FQA C35H31O15 691.1668 691.1647 3.1

64 3-O-feruloyl-4,5-di-O-caffeoylquinic acid 3F-4,5-diCQA C35H31O15 691.1668 691.1711 -6.2a

65 3,4-di-O-caffeoyl-5-O-feruloylquinic acid 3,4-diC-5FQA C35H31O15 691.1668 691.1647 3.1

66 3-O-caffeoyl-4,5-di-O-feruloylquinic acid 3C-4,5-diFQA C36H33O15 705.1825 705.1851 -3.8

67 3,4-di-O-feruloyl-5-O-caffeoylquinic acid 3,4-diF-5CQA C36H33O15 705.1825 705.1833 -1.1

68 3,4-di-O-caffeoyl-5-O-sinapoylquinic acid 3,4-diC-5SiQA C36H33O16 721.1774 721.1795 -2.9

69 3-O-sinapoyl-4,5-di-O-caffeoylquinic acid 3Si-4,5-diCQA C36H33O16 721.1774 721.1766 1.1

a In 10 chromatographic runs it was the lowest error and reason for this higher valuemight be lower concentration of the compound in extract. All MSn data were in agreement onthe presence of compound 64.


Robusta beans analyzed with some noticeable differences inintensities. A detailed discussion about these differences is outsidethe scope of this paper and requires a full principal componentanalysis (see the Supporting Information for a listing of theCGAsfound in the Robusta beans).

In a second round of experiments using the same chromato-graphic conditions, tandem MS spectra were recorded in thenegative ion mode using an ion trap MS detector. Selected ionmonitoring and analysis of the individual tandem MS spectraagain revealed and confirmed the presence of those 37 chloro-genic acids in all six green Robusta coffee bean samples.

In general, new CGAs can be identified in an All MSn EIC(extracted ion chromatogram) by their unique fragments at m/z173 and 191. Selected ion monitoring at m/z 397, 559, 573, 588,and 719 immediately located 16 chromatographic peaks elutingbetween 23 and 60 min, each with a UV spectrum typical ofchlorogenic acids (λmax 320 nm). However, because the quinicacid moiety is not acylated at C1 in coffee beans and producesonly three caffeoylquinic acids (1-3), three feruloylquinic acids(4-6), and three p-coumaroylquinic acid (7-9), only threesinapoylquinic acids (13-15) and three trimethoxycinnamoyl-quinic acids (72-74) were expected. During this study we did notobserve any monoacyl trimethoxycinnamoylquinic acid isomers.Three peaks produced MS2 fragment ions at either m/z 173 or191, 12 peaks producedMS3 fragment ions atm/z 173 or 191, anda further peak produced MS4 fragment ions at m/z 173 or 191consistent with the presence of a quinic acid residue (5), and theabsence of anMS3 fragment ion atm/z 205, 219, or 233 confirmedthat none of these substance were derivative of alkyl quinate. Fiveof the 16 peaks produced MS2 ions at m/z 397 ([sinapoylquinicacid-Hþ]-) and 379 ([sinapoylquinic acid-H2O-Hþ]-) andMS3 ions at m/z 223 ([sinapic acid - Hþ]-) analogous to thoseproduced by dicaffeoylquinic acids, caffeoyl-feruloylquinic acids,dimethoxycinnamoyl-feruloylquinic acid, and p-coumaroyl-feruloylquinic acids (5 , 7 , 8 , 12 ).

The other two peaks producedMS2 ions atm/z 353 and 349 andMS3 ions atm/z 173 and 193. TheMS2 ion atm/z 353 and theMS3

ion at m/z 173 are characteristic (5) of caffeic acid derived diacylchlorogenic acids and can be assigned as [caffeoylquinic acid- Hþ]-, and the MS2 ion at m/z 349 and MS3 ion at m/z 193 arecharacteristic of ferulic acid derived diacyl chlorogenic acids andcan be assigned as [sinapoylquinic acid-H2O- CH2O-Hþ]-.The remaining ions were tentatively assigned to a similar series offragments 30 amu larger than the ferulic acid related fragments (46amu larger than the caffeic acid related fragments), stronglysuggesting that they might have one more methyl ether group atposition 5 of the cinnamic acid residue, that is, 5-O-methyl ether offerulic acid (sinapic acid). Those compounds ofMr 574 producedMS2 ions at m/z 379 and compounds of Mr 560 produced MS2

ions at m/z 349 were tentatively assigned as feruloyl-sinapoylqui-nic acids and caffeoyl-sinapoylquinic acids, respectively.

Another five peaks produced MS2 ions at m/z 411([trimethoxycinnamoylquinic acid - Hþ]-) and MS3 ions at m/z 237 ([trimethoxycinnamic acid - Hþ]-), analogous to thoseproduced by caffeoyl-feruloylquinic acids, dimethoxycinnamicacids, and p-coumaroyl-feruloylquinic acids (5, 7, 8, 12). Theother two peaks producedMS2 ion atm/z 349 andMS3 ion atm/z193. The MS2 ion at m/z 349 and the MS3 ion at m/z 193 arecharacteristic (5) of ferulic acid derived diacyl chlorogenic acidsand can be assigned as [feruloylquinic acid - H2O - Hþ]-. Theremaining ions were tentatively assigned to similar series offragments 30 amu larger than the dimethoxycinnamic acid,strongly suggesting that they might have one more methyl ethergroup at position 5 of the cinnamic acid, that is, 5-O-methyl etherof 3,4-dimethoxycinnamic acid (trimethoxycinnamic acid). Thosecompounds of Mr 573 produced MS2 ions at m/z 393, 353, and335 and compounds ofMr 587 producedMS2 ion atm/z 349 weretentatively assigned as caffeoyl-trimethoxycinnamoylquinic acidsand feruloyl-trimethoxycinnamic acids, respectively. It is conceiv-able that a feruloyl-sinapoylquinic acid, a caffeoyl-sinapoylquinicacid, and a feruloyl-trimethoxycinnamic acid might produce ionsatm/z 379 or 349, and there can be no doubt that all three series ofchlorogenic acids were present.

Characterization of Putative Sinapoylquinic Acids (Mr 398).TheRobusta extract contained three minor components with mole-cular ions atm/z 397 that eluted between 23 and 30min. All threepeaks had UV spectra typical of chlorogenic acids. MS2 data arepresented in Table 2.

Because the monoacyl chlorogenic acids examined on diphenylpacking elute in the sequence 3-acyl, 5-acyl, and 4-acyl, the firstsinapoylquinic acid was tentatively assigned as the 3-isomer (13).TheMS2 base peak atm/z 223 ([sinapoylquinic acid- quinic acid-Hþ]-), which subsequently decarboxylates and demethylates atMS3 (concentration was below the limits of detection but com-pared with the MS4 of diacyl CGAs), is analogous behavior to3-feruloylquinic acid (4) and 3-dimethoxycinnamoylquinic acid(10) (5,7,8). Themost strongly retained of the sinapoylquinic acidisomers has a fragmentation characteristic of a 4-acyl chlorogenicacid (MS2 and MS3 base peaks at m/z 173 and 93, respectively,which were not in the range of the limit of detection) and can betentatively assigned as 4-sinapoylquinic acid (14). The thirdisomer has fragmentation characteristic of a 5-acyl chlorogenicacid (MS2 at m/z 191) and the retention time is between those of3- and 4-acylated sinapoylquinic acid, which strongly suggest5-sinapoylquinic acid (15). Further evidence for the assignment ofmonoacyl sinapoylquinic acids (13-15) came from an indepen-dent synthesis of a mixture of all four possible regioisomersof sinapoylquinic acid (Figure 2). A direct comparison of thechromatograms of the synthetic mixture and coffee bean extractsallows an ambiguous confirmation of the presence of sinapoyl-quinic acids (13-15) by comparison of retention times, UV-visdata, and tandem mass spectra (Figure 3 and Table 2).

Table 2. Negative Ion MS2 and MS3 Fragmentation Data for Monoacyl Sinapoylquinic Acids (13-15 and 70) and Trimethoxycinnamoylquinic Acids (71-74)

MS1 MS2 MS3

base peak secondary peak base peak secondary peak

no. compd parent ion m/z m/z int m/z int m/z m/z int m/z int m/z int

13 3-SiQA 397.0 222.9 164.0 13 149.0 10 163.9 178.9 15 148.9 15

14 4-SiQA 397.0 172.9 222.9 12 93.3 111.1 62 71.6 58 154.9 22

15 5-SiQA 397.0 190.9 222.9 5 127.0 172.9 70 93.3 60 85.4 59

70 1-SiQA 397.0 222.9 172.9 21 190.7 7 207.9 178.9 32 163.9 31

71 1-TQA 411.1 172.9 143.0 20 236.9 10 143.0 155.0 50 111.2 60 93.3 43

72 3-TQA 411.0 236.9 133.0 221.9 50 192.9 40 103.3 23

73 5-TQA 411.0 172.9 93.0 155.0 20 111.1 45 61.8 5

74 4-TQA 411.1 172.9 236.9 21 93.0 155.0 15 111.1 49 71.6 13


Characterization of Putative Caffeoyl-sinapoylquinic Acids (Mr

560).Of the four peaks that yieldedmolecular ions atm/z 559, twoproduced anMS2 base peak atm/z 397 with anMS3 base peak atm/z 223, suggestive of caffeoyl-sinapoylquinic acids. The thirdpeak showed an MS2 base peak at m/z 349 followed by an MS3

base peak at m/z 192.8, and the fourth peak produced an MS2

base peak at m/z 353. Theoretically, six caffeoyl-sinapoylquinicacid isomers would have been expected, as seen previously (5) forthe caffeoyl-feruloylquinic acids. Table 3 compares the fragmen-tation patterns of the three putative caffeoyl-sinapoylquinic acidswith data previously obtained for the six caffeoyl-feruloylquinicacids (5). The caffeoyl-feruloylquinic acids were assigned by

comparing their fragmentation behavior at MS(nþ1) with theMSn fragmentation of caffeoylquinic acids and feruloylquinicacids and the MS(nþ1) fragmentation of the dicaffeoylquinicacids (5 ). For chlorogenic acids not substituted at C1, it wasobserved that a caffeoyl at C5 was the most easily eliminated,whereas that at C4 was the most stable and that at C3 was ofintermediate stability. For five of the six caffeoyl-feruloylquinicacids this resulted at MS2 in one of the two cinnamoyl residuesbeing eliminated almost exclusively (maximally 10% loss of thesecond cinnamoyl residue in 3-caffeoyl-4-feruloylquinic acid). Incontrast, 3-feruloyl-4-caffeoylquinic acid lost both cinnamoylresidues with nearly equal facility, giving an MS2 base peak

Figure 2. Synthesis of mixture of regioisomers of sinapoylquinic acid (13-15 and 70).

Figure 3. Extracted ion chromatogram (EIC) at m/z 397 (compounds 13-15 and 70) in negative ion mode.


(m/z 353) accompanied by a very intense (90% of base peak)secondary ion atm/z 367. Significant amounts of the correspond-ing dehydrated ions (m/z 335 and 349) were also produced (5).

By analogy with the relative hydrophobicity of the caffeoyl-feruloylquinic acids and dicaffeoylquinic acids (5,12,13) it shouldbe expected that the first caffeoyl-sinapoylquinic acid to elutewould be one of the 3,4-caffeoyl-sinapoylquinic acid isomers.However, we observed earliest elution of 3-sinapoyl-5-caffeoyl-quinic acid, which loses its caffeoyl residue before its sinapoylresidue. It has been shown that a caffeoyl residue at C5 is easilyeliminated (5). Such an elimination would produce [3-sinapoyl-quinic acid-Hþ]- as the MS2 base peak, and as a argued abovefor 49, the MS3 base peak at m/z 223 is consistent with such anassignment, suggesting that the first eluting caffeoyl-sinapoylqui-nic acid is 3-sinapoyl-5-caffeoylquinic acid (49). Its fragmentationpattern more closely resembles 3-feruloyl-5-caffeoylquinic acid(27) than 3-caffeoyl-5-feruloylquinic acid (28).

The next eluting caffeoyl-sinapoylquinic acid loses its twocinnamoyl residues with nearly equal facility (slightly morefavoring the sinapoyl residue) and produces strong dehydratedions at m/z 335 and 379, thus closely resembling 3-feruloyl-4-caffeoylquinic acid (25) (Table 3) (5). Its MS3 spectrum was notdetected and tentatively assigned as 3-sinapoyl-4-caffeoylquinicacid (50).

The fragmentation of the third eluting isomer (51) produced anMS2 base peak at m/z 349, an MS3 base peak at m/z 193, and anMS4 base peak at m/z 134, which we consider fingerprint ions offeruloylquinic acids. AnMS3 secondary peak atm/z 173 (80% ofbase peak), which is consistent with a 4-substituted chlorogenicacid, was observed (8). The MS2 and MS3 spectra indicate thepresence of a feruloyl substituent at C4 rather than a caffeoyl orsinapoyl substituent (8). Two secondary peaks atm/z 269 and 305are characteristic for a diacyl CGA containing a feruloyl sub-stituent at C4, that is, 3,4-acylated CGA (8). The MS2 base peakat m/z 349 ([diacyl CGA - H2O - cinnamoyl - Hþ]-) is due tothe loss of H2O (18 amu) and one cinnamoyl residue (192 amu).This identifies the second acyl group as a methoxy-dihydroxycinnamic acid residue. There are two possible regiochemistries toconsider, 4-methoxy A or 3-methoxy B. In the absence of furtherevidence in favor of either structure A or B, structure B was con-sideredmore likely because it would represent a logical biosynthetic

precursor for 3Si-4FQA, and 51 was tentatively assigned as4-feruloyl-3-(3,4-dihydroxy-3-methoxy)cinnamoylquinic acid. Tothe best of our knowledge this hydroxycinnamoyl substituent isreported here for the first time in nature.

The most hydrophobic isomer has the characteristic fragmen-tation of a vic-diacyl-chlorogenic acid (MS3 and MS4 base peaksat m/z 173 and 93, respectively), and it clearly loses its caffeoylresidue before its sinapoyl residue [MS2 base peak atm/z 397 anda strong MS3 secondary ion (99% of base peak) atm/z 223]. TheMS2 base peak must be either [4-sinapoylquinic acid - Hþ]- or[5-sinapoylquinic acid - Hþ]-, and the subsequent fragmenta-tion to the dehydrated ion m/z 173 indicates [4-sinapoylquinicacid - Hþ]- and is thus tentatively assigned as 3-caffeoyl-4-sinapoylquinic acid (52). The absence of the MS2 ions atm/z 299and 255, suggesting that 52 does not have a caffeoyl residue atC4,is consistent. Its fragmentation pattern more closely resembles3-caffeoyl-4-feruloylquinic acid (26) than 3-feruloyl-4-caffeoyl-quinic acid (25) (Table 3).

Characterization of Putative Feruloyl-sinapoylquinic Acids (Mr

574). The LC-MS data for the feruloyl-sinapoylquinic acids(53-55) are summarized in Table 3. The fragmentation patternof the slowest eluting isomer (55) is identical to that of caffeoyl-sinapoylquinic acid 52 if allowance is made for the weak caffeicacid-derived ion at m/z 335 being replaced by the analogous, butmore intense, m/z 349, suggesting that logically it can be tenta-tively assigned as 4-sinapoyl-3-feruloylquinic acid (55).

The preceding feruloyl-sinapoylquinic acid 54 resembles theanalogous feruloyl-dimethoxycinnamoylquinic acid 34 (Table 3)in that both produce a dehydratedMS2 base peak. On the basis ofthe above argument, feruloyl-sinapoylquinic acid was tentativelyassigned as 4-sinapoyl-5-feruloylquinic acid (54).

The fragmentation of the most rapidly eluting feruloyl-sina-poylquinic acid (53) resembles the analogous caffeoyl-sinapoyl-quinic acid (49) in that both lose their caffeoyl or feruloyl residuemore preferentially than their sinapoyl residue. Fragmentation ofthe MS2 base peak (m/z 397) produces an MS3 base peak at m/z223 (Table 3), which is consistentwith 3,5-caffeoyl-sinapoylquinicacid. Accordingly, it is assigned as 3-sinapoyl-5-feruloylquinicacid (53).

From our previous studies it has been proven thatMSn spectraofmonoacyl CGAs are similar toMSnþ1 spectra of corresponding

Table 3. Negative Ion MS2, MS3, and MS4 Fragmentation Data for the Dimethoxycinnamoyl-caffeoylquinic Acids, Dimethoxycinnamoyl-feruloylquinic Acid, Caffeoyl-sinapoylquinic Acids, Feruloyl-sinapoylquinic Acids, Trimethoxycinnamoyl-caffeoylquinic Acids, and Trimethoxycinnamoyl-feruloylquinic Acids

MS1 MS2 MS3 MS4

base peak secondary peak base peak secondary peak base peak secondary peak

no. compd parent ion m/z m/z int m/z int m/z int m/z m/z int m/z int m/z int m/z m/z int m/z int

31 3D-4CQA 543.0 381.1 335.1 92 363.1 4 298.7 12 207.1 173.3 10 149.1

32 3D-5CQA 543.0 381.1 335.1 2 207.1 173.3 25 149.0 162.7 80 133.1 22

33 4D-5CQA 543.0 381.0 335.1 2 173.1 207.0 55 191.0 32 135.1 9 93.2 155.1 20 111.1 35

34 3D-4FQA 557.0 349.1 363.1 75 172.9 178.9 76 190.9 9 135.0 19 93.1 111.0 20

35 3D-5FQA 557.0 381.2 349.2 75 175.0 193.0 80 269.0 55 313.0 25 160.1

36 4D-5FQA 557.0 381.2 349.2 30 207.1 173.3 47 93.1 155.1 10 111.1 40

49 3Si-5CQA 559.1 397.1 335.2 5 375.0 20 222.8 163.7 148.9 20

50 3Si-4CQA 559.1 353.1 334.9 35 378.9 28 397.0 60

51 3C-4SiQA 559.1 349.0 333.0 16 490.7 20 521.0 42 192.8 172.6 45 149.0 60 154.6 50 93.2 111.1 40

52 4Si-5CQA 559.1 397.1 172.9 222.8 99 149.0 5 98.9 83.2 36

53 3Si-5FQA 573.1 397.0 349.1 40 222.8 163.8 148.7 36

54 3F-4SiQA 573.1 379.0 349.1 60 367.0 18 397.0 40 222.8 204.8 90 166.8 38 283.8 60

55 4Si-5FQA 573.1 397.1 349.1 8 379.0 5 172.7 222.8 60 93.2 155.0 12 111.0 10

56 4T-5CQA 573.1 411.1 236.9 10 434.8 55 502.7 172.8 236.8 70

57 3T-5CQA 573.1 411.0 236.9 30 379.0 5 535.1 8 236.8 172.8 20 132.8 221.8 45

58 3T-5FQA 587.1 411.0 348.9 65 236.9 20 172.9 10 236.9 172.9 10 133.1

59 3T-4FQA 587.2 348.9 411.1 45 393.0 55 236.9 65 192.9 268.9 20 172.9 42

60 4T-5FQA 587.2 411.0 348.9 30 236.9 17 172.9 12 93.0 111.0 62


diacyl CGAs. For the further identification and characterizationof trimethoxycinnamic acid containing diacyl CGAs, a nonselec-tive synthesis of a mixture of all four possible regioisomers ofmonoacyl trimethoxycinnamoylquinic acid (63-66) has beencarried out (Figure 4). In the first step of synthesis, trimethoxy-cinnamic acid was converted into its acid chloride by using oxalylchloride and DMF to give the acid chloride. This was sub-sequently reacted with quinic acid in the presence of pyridineandDMAP to yield amixture of four regioisomers of trimethoxy-cinnamoylquinic acid (63-66). A fragmentation pattern and LCbehavior of all four regioisomers of trimethoxycinnamoylquinicacid is discussed below.

LC-MSnCharacterization of Trimethoxycinnamoylquinic Acid

Regioisomers (71-74). The monoacyl trimethoxycinnamic acidsexaminedondiphenyl packing elute in the sequence 1-acyl, 3-acyl,5-acyl, and 4-acyl trimethoxycinnamic acid (Figure 5). The second

eluting isomer 72 produced an MS2 base peak at m/z 237([trimethoxycinnamic acid - Hþ]-), which subsequently decar-boxylates and demethylates at MS3 (m/z 133) (Figure 6 andTable 2) and is analogous behavior to 3-feruloylquinic acid (4)and 3-dimethoxycinnamic acid (10) (5, 7, 8). All three otherisomers producedMS2 base peaks atm/z 173 ([quinic acid-H2O- Hþ]-), and MS2 secondary peaks at m/z 237 ([trimethoxycin-namic acid - Hþ]-) (which is absent in 5-trimethoxycinnamoyl-quinic acid 73). 4- (74) and 5-trimethoxycinnamic acid producedMS3 base peaks atm/z 93, and 1-trimethoxycinnamoylquinic acid71 produced MS3 base peak at m/z 143.

Characterization of Putative Caffeoyl-trimethoxycinnamoylqui-

nic Acids (Mr 574). Two peaks were detected atm/z 573, and theirLC-MS data are summarized in Table 3. The first eluting isomer56 produced anMS2 base peak atm/z 411 by the loss of a caffeoylresidue ([trimethoxycinnamoylquinic acid-Hþ]-) andMS3 base

Figure 4. Synthesis of regioisomers of trimethoxycinnamoylquinic acid (71-74).

Figure 5. Extracted ion chromatogram (EIC) of synthetic trimethoxycinnamoylquinic acids at m/z 411 (negative ion mode).


peak at m/z 172.8 and MS2 secondary peak at m/z 236.9([trimethoxycinnamic acid - Hþ]-). MS2 and MS3 data of 56are similar toMS2 data of 4-trimethoxycinnamoylquinic acid 74,which confirms the presence of a trimethoxycinnamoyl residue atC-4 of the quinic acid moiety. The absence of an MS2 second-ary peak at m/z 353 ([caffeoylquinic acid - Hþ]-) or 335([caffeoylquinic acid-H2O-Hþ]-) confirms the caffeoyl residueat C-5, and isomer 56 was tentatively assigned as 5-caffeoyl-4-trimethoxycinnamoylquinic acid. The next eluting isomer, 57,produced an MS2 base peak at m/z 411 by the loss of a caffeoylresidue ([trimethoxycinnamoylquinic acid - Hþ]-), an MS3 basepeak at m/z 236.8, and an MS4 base peak at m/z 132.8. MS3 andMS4 spectra of 57 are similar to MS2 and MS3 spectra of3-trimethoxycinnamoylquinic acid (72), which confirms the pre-sence of a trimethoxycinnamoyl residue atC3. The caffeoyl residueshould be atC5because if it is atC4, then theMS3base peakwill beatm/z 173 instead of 236.8. On the basis of the above arguments 57was tentatively assigned as 3-trimethoxy-5-caffeoylquinic acid.

Characterization of Putative Feruloyl-trimethoxycinnamoylqui-

nic Acids (Mr 588). The LC-MS data for the feruloyl-trimethoxy-cinnamoylquinic acids (58-60) are summarized in Table 3. Thefragmentation of the slowest eluting isomer (60) is identical tothat of feruloyl-sinapoylquinic acid 55 if allowance ismade for theweak ferulic acid-derived ion at m/z 349 being replaced by themore intense peak, suggesting that logically it can be tentativelyassigned as 4-trimethoxycinnamoyl-5-feruloylquinic acid.

The preceding feruloyl-trimethoxycinnamoylquinic acid 59

resembles the analogous feruloyl-sinapoylquinic acid 54 (Table 3)

in that both produce adehydratedMS2base peak.On the basis of theabove argument, feruloyl-trimethoxycinnamoylquinic acidwas tenta-tively assigned as 3-trimethoxycinnamoyl-4-feruloylquinic acid (59).

The fragmentation of the most rapidly eluting feruloyl-tri-methoxycinnamoylquinic acid 58 resembles the analogous feru-loyl-sinapoylquinic acid 53 in that both lose feruloyl residuemorepreferentially than sinapoyl or trimethoxycinnamoyl residue.Fragmentation of the MS2 base peak (m/z 411) produces anMS3 base peak at m/z 237 (Table 3), which is consistent with3-sinapyl-5-feruloylquinic acid. Accordingly, it is assigned as3-trimethoxycinnamoyl-5-feruloylquinic acid.

Characterization of Putative Triacylquinic Acids. Next to themonoacyl and diacyl quinic acid derivatives a series of triacylquinic acids derivatives were identified by their characteristicquinic acid fragments atm/z 191 and 173 in theMS4 tandemmassspectra. According to our hierarchical scheme for assignment ofchlorogenic acid regiochemistry (5-8), the following assumptionswere made to assign the regiochemistry of the triacyl quinic acidderivatives.

(1) All of the CGAs reported so far are not acylated at C-1 ofthe quinic acid moiety, which suggests the absence of C-1 acyla-tion, so we consider only the presence of 3-,4-, and 5-acyl CGA.

(2) Loss of the acyl moiety occurs most readily from C-5.Hence, the fragments and neutral losses inMS2 define the natureof the substituent at C-5.

(3) Loss of the acyl moiety occurs more reluctantly from C-3.Hence, the fragments and neutral losses inMS3 define the natureof the substituent at C-3.

Figure 6. Fragmentation mechanism of trimethoxycinnamoylquinic acids (71-74).


(4) Loss of the acyl moiety occurs most reluctantly accompa-nied by loss of water from C-4. Hence, the fragments and neutrallosses in MS4 define the nature of the substituent at C-4.

Following identification of a total of nine triacyl quinic acidsby all MSn searches, the molecular formulas were confirmed(Table 1) according to their high-resolution mass data in thenegative ion mode. The mass error was typically under 5 ppm,confirming the molecular formulas of the triacyl quinic acids.Selected ion monitoring at m/z 677, 691, 705, 719, and 721immediately located nine chromatographic peaks eluting between40 and 60 min, each with a UV spectrum typical of chlorogenicacids (λmax 320 nm). Structures have been assigned on the basis ofLC-MSn patterns of fragmentation, and MSn data are given inTable 4. This detailed analysis ofMSn data allowed the identifica-tion of 3-dimethoxycinnamoyl-4-feruloyl-5-caffeoylquinic acid(61; Mr 720); 3,4,5-tricaffeoylquinic acid (62; Mr 678); 3,5-dicaffeoyl-4-feruloylquinic acid (63); 3-feruloyl-4,5-dicaffeoyl-quinic acid (64), 3,4-dicaffeoyl-5-feruloylquinic acid (65; Mr

692); 3-caffeoyl-4,5-diferuloylquinic acid (66), 3,4-diferuloyl-5-caffeoylquinic acid (67; Mr 706); 3,4-dicaffeoyl-5-sinapoylquinicacid (68); and 3-sinapoyl-4,5-dicaffeoylquinic acid (69; Mr 722).Compounds 61 and 63-69 have been reported here for the firsttime in nature and for the first time from this source.

Missing Isomers. Because all six theoretically possible regio-isomers of caffeoyl-feruloylquinic acids (25-30) are easily locatedin the Robusta coffee extract (5), it is a little surprising that onlythree caffeoyl-sinapoyl (49, 50, and 52), three feruloyl-sinapoyl-quinic acids (53-55), two caffeoyl-trimethoxycinnamoylquinicacids (56 and 57), and three feruloyl-trimethoxycinnamoylquinicacids (58-60) have been observed. This could reflect the failure ofthe coffee plant to synthesize all six isomers. However, all three

sinapoylquinic acids are produced, which could be elaboratedinto the full set of diacyl derivatives. It is possible that these otherisomers are present but below the limits of detection in ourmethod.If that is the case, then the most concentrated of the “missing”caffeoyl-sinapoylquinic acids, feruloyl-sinapoylquinic acids, caffeoyl-trimethoxycinnamoylquinic acids, and feruloyl-trimethoxy-cinnamoylquinic acids must be present at concentrations con-siderably below that of the weakest isomer found, because analysisof deliberately (10 times) concentrated extracts did not allow theirdetection. A third possibility is that they are present at low con-centration and not resolved chromatographically.

Intermediates in Biosynthesis of Trimethoxycinnamoyl Quinic

Acid Derivatives. From the current study an interesting detailbecomes apparent. For one full series of compounds (3-feruloyl4-acyl quinic acids) a full series of possible biochemical inter-mediates have been observed, hence allowing the proposal of abiosynthetic route for the biosynthesis of trimethoxycinnamoylquinic acids. This proposed route is based on the structures offour intermediates observed and shown in Figure 7. We proposethat starting from 3,4-diferuloyl quinic acid (19) an oxidaseenzyme inserts an oxygen into position 5 of the 4-feruloylsubstituent, yielding intermediate 51. Methylation presumablyby catechol-O-methyl transferase (COMT) yields 3-feruloyl-4-sinapoyl quinic acid (55). A final COMT-mediated methylationfurnishes 3-ferulyol-4-trimethoxycinnamoyl quinic acid (59).

Conclusion.All sinapoylquinic acid and trimethoxycinnamoyl-quinic acid derivatives could be located in all 6 Robusta samples,whereas theywere absent in 10Arabica samples investigated.As apreliminary conclusion, we therefore suggest that sinapoyl quinicacid derivatives can act as phytochemical markers that enablethe distinction between Arabica and Robusta coffee varieties.

Table 4. Negative Ion MS2, MS3, and MS4 Fragmentation Data for the Triacylquinic Acids 61-69

MS1 MS2 MS3 MS4

base peak secondary peak base peak secondary peak base peak secondary peak

no. compd parent ion m/z m/z int m/z int m/z m/z int m/z int m/z int m/z m/z int m/z int

61 3D-4F-5CQA 719.3 557.1 349.1 5 349.0 381.0 25 304.9 5 268.8 10 174.7 192.9 85 149.0 25

62 3,4,5-triCQA 677.1 515.1 497.0 5 353.1 25 353.0 335.0 10 254.8 6 298.9 6 172.7 178.9 80 190.8 50

63 3,5diC-4FQA 691.1 529.1 335.0 28 367.0 19 367.0 334.9 15 172.7 55 172.7 192.7 40

64 3F-4,5-diCQA 691.2 529.0 515.1 30 352.9 367.1 55 254.8 5 172.7 178.8 80 190.8 22

65 3,4-diC-5FQA 691.2 515.1 529.1 90 497.0 10 352.9 366.9 50 334.9 8 254.7 8 172.7 178.7 80 190.8 50

66 3C-4,5-diFQA 705.2 529.1 543.0 72 511.0 23 367.0 353.0 74 335.0 37 172.6 192.7 16

67 3,4-diF-5CQA 705.2 543.0 529.0 50 349.0 367.0 42 381.0 20 172.7 26 192.7 268.7 53 312.9 82

68 3,4-diC-5SiQA 721.2 515.1 559.1 59 352.9 335.0 25 298.8 5 255.0 2 172.6 178.6 80 191.0 50

69 3Si-4,5-diCQA 721.2 559.1 515.1 23 298.9 8 353.0 335.0 10 172.7 26 172.7 178.6 35 190.6 70

Figure 7. Proposed scheme for the biosynthesis of trimethoxycinnamoylquinic acids (COMT = catechol-O-methyl transferase).


Interestingly, differences in such minor components have not yetbeen established in other works studying the phytochemicalprofile of coffee varieties, mainly by multivariant statistical tech-niques (14, 15). Detailed knowledge of these differences mighthelp to distinguish adulterations of ground or roasted coffeeproducts in the future. Whether these classes of secondary meta-bolites contribute to the inferior sensory properties of Robustacoffee must be the topic of further investigations. As in other ex-amples reported in the literature, the chlorogenic acid profile ofindividual plant species or variety can serve as a phytochemicalprofile of the species investigated (16-18).

ACKNOWLEDGMENT

Excellent technical support by Anja M€uller is acknowledged.

Supporting Information Available: Additional figures and

table. This material is available free of charge via the Internet at

http://pubs.acs.org.

LITERATURE CITED

(1) IUPAC. Nomenclature of cyclitols. Biochem. J. 1976, 153, 23-31.(2) Clifford, M. N. Chlorogenic acids and other cinnamates - nature,

occurrence, dietary burden, absorption and metabolism. J. Sci. FoodAgric. 2000, 80, 1033–1043.

(3) Clifford, M. N. Chlorogenic acids and other cinnamates - nature,occurrence and dietary burden. J. Sci. Food Agric. 1999, 79, 362–372.

(4) Eliel, E. L.; Ramirez, M. B. (-)-Quinic acid: configurational(stereochemical) descriptors. Tetrahedron: Asymm. 1997, 8, 3551–3554.

(5) Clifford, M. N.; Johnston, K. L.; Knight, S.; Kuhnert, N. Hier-archical scheme for LC-MSn identification of chlorogenic acids.J. Agric. Food Chem. 2003, 51, 2900–2911.

(6) Clifford, M. N.; Jarvis, T. The chlorogenic acids content of greenrobusta coffee beans as a possible index of geographic origin. FoodChem. 1988, 29, 291–298.

(7) Clifford,M.N.;Marks, S.; Knight, S.; Kuhnert, N. Characterizationby LC-MSn of four new classes of p-coumaric acid-containing diacyl

chlorogenic acids in green coffee beans. J. Agric. Food Chem. 2006,54, 4095–4101.

(8) Clifford, M. N.; Knight, S.; Surucu, B.; Kuhnert, N. Characteriza-tion by LC-MSn of four new classes of chlorogenic acids in greencoffee beans: dimethoxycinnamoylquinic acids, diferuloylquinicacids, caffeoyl-dimethoxycinnamoylquinic acids, and feruloyl-dimethoxycinnamoylquinic acids. J. Agric. Food Chem. 2006, 54,1957–1969.

(9) Frank, O.; Blumberg, S.; Krumpel, G.; Hofmann, T. Structuredetermination of 3-O-caffeoyl-epi-γ-quinide, an orphan bitter lac-tone in roasted coffee. J. Agric. Food Chem. 2008, 56, 9581–9585.

(10) Clifford, M. N.; Knight, S. The cinnamoyl-amino acid conjugates ofgreen robusta coffee beans. Food Chem. 2004, 87, 457–463.

(11) Egan, H.; Kirk, R. S.; Sawyer, R. Pearson’s Chemical Analysis ofFoods; Churchill Livingstone: London, U.K., 1981.

(12) Clifford, M. N.; Knight, S.; Kuhnert, N. Discriminating between thesix isomers of dicaffeoylquinic acid byLC-MSn. J. Agric. FoodChem.2005, 53, 3821–3832.

(13) Clifford, M. N. The analysis and characterisation of chlorogenicacids and other cinnamates. In Methods in Polyphenol Analysis;Santos-Buelga, C., Williamson, G., Eds.; Royal Society of Chemistry:Cambridge, U.K., 2003; Vol. 14, pp 314-337.

(14) Alonso-Salces, R.M.; Serra, F.; Reniero, F.; Heberger, K. Botanicaland geographical characterization of green coffee (Coffea arabicaand Coffea canephora): chemometric evaluation of phenolic andmethylxanthine contents. J. Agric. Food Chem. 2009, 53, 4224–4235.

(15) Mendonca, J. C. F.; Franca, A. S.; Oliveira, L. S.; Nunes, M.Chemical characterisation of non-defective and defective greenArabica and Robusta coffees by electrospray ionization-massspectrometry (ESI-MS). Food Chem. 2008, 111, 490–497.

(16) Clifford, M. N.; Zheng, W.; Kuhnert, N. Profiling the chlorogenicacids of aster by HPLC-MSn. Phytochem. Anal. 2006, 17, 384–393.

(17) Clifford, M. N.; Wu, W. G.; Kuhnert, N. The chlorogenic acids ofHemerocallis. Food Chem. 2006, 95, 574–578.

(18) Clifford, M. N.; Stoupi, S.; Kuhnert, N. The LC-MSn analysis of cisisomers of chlorogenic acids. Food Chem. 2008, 106, 379–385.

Received for review April 16, 2010. Revised manuscript received June

29, 2010. Accepted July 6, 2010. Financial support from Jacobs

University Bremen is gratefully acknowledged.

Profile and Characterization of the Chlorogenic Acids in Green Robusta Coffee Beans by LCMS n :...

Documents

Transcript of Profile and Characterization of the Chlorogenic Acids in Green Robusta Coffee Beans by LCMS n :...